Classification of Flood Levels using Random Forest and Support Vector Machine Algorithms

Abstract

Floods are one of the most common natural disasters in Indonesia. This study analyzes the impact of each flood event by examining factors such as duration, water level, and the number of affected individuals to identify flood characteristics based on severity. Climate variables such as temperature, humidity, rainfall, and wind speed were investigated as parameters characterizing flood occurrences. The primary objective of this research is to classify flood levels using Random Forest and Support Vector Machine (SVM) algorithms, and to evaluate the accuracy of these classifications using a Confusion Matrix. The outcomes are intended to inform decision-making processes during floods, thereby aiming to minimize associated losses. The research utilized historical flood data from the DKI Jakarta BPBD, accessed through the Satu Data Jakarta website, and climate data from the BMKG Geophysical Station, covering the period from 2013 to 2020. The Random Forest classification system demonstrated exceptional performance, achieving an accuracy of 99.21%. Similarly, the SVM classification system performed robustly, with an accuracy of 98.43%. Both models initially exhibited overfitting during the early stages of model development. However, this issue is diminished as the dataset size increases, thereby enhancing the models' generalization capabilities.